2012-05-18T12:41:06+02:00 http://cvpapers.com/forum/ Lussumo Vanilla & Feed Publisher http://cvpapers.com/forum/comments.php?DiscussionID=11&Focus=35#Comment_35 2012-04-14T10:49:33+02:00 2012-05-18T12:41:06+02:00 loulouRecognition http://cvpapers.com/forum/account.php?u=61

I have a problem please help me. I am developping a C++ application for face authentification.So the first steps to do are:face detection and preprocessing face images. For face detection iI have ...

For face detection iI have used the HaarCascadeClassifier:theproblem is that the this tool or this algorithm gives me a facial region detected by a little bit large rectangle that englobes hair and some of the background.Is there a solution to chanage the dimension of this rectangle (i used frontalfacecascaadclassifier)
for face preprocessing i want to do face alignment exactly like this technique http://www.bytefish.de/blog/aligning_face_images but but don't know how? do i need to change my detectcion technique.
I want to obtain an image like this : http://www.bytefish.de/_media/blog/arnie_20_20_200_200.jpg?cache=&w=200&h=200 but i have obtained an image like this http://www.bytefish.de/_media/blog/arnie_20_20_200_200.jpg?cache=&w=200&h=200.
Thanks in advance for your help]]> http://cvpapers.com/forum/comments.php?DiscussionID=10&Focus=34#Comment_34 2012-04-14T10:46:29+02:00 2012-05-18T12:41:06+02:00 SSSSS http://cvpapers.com/forum/account.php?u=60

I have a problem please help me. I am developping a C++ application for face authentification.So the first steps to do are:face detection and preprocessing face images. For face detection iI have ...

For face detection iI have used the HaarCascadeClassifier:theproblem is that the this tool or this algorithm gives me a facial region detected by a little bit large rectangle that englobes hair and some of the background.Is there a solution to chanage the dimension of this rectangle (i used frontalfacecascaadclassifier)
for face preprocessing i want to do face alignment exactly like this technique http://www.bytefish.de/blog/aligning_face_images but but don't know how? do i need to change my detectcion technique.
I want to obtain an image like this : http://www.bytefish.de/_media/blog/arnie_20_20_200_200.jpg?cache=&w=200&h=200 but i have obtained an image like this http://www.bytefish.de/_media/blog/arnie_20_20_200_200.jpg?cache=&w=200&h=200.
Thanks in advance for your help]]> http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=33#Comment_33 2011-12-28T09:41:23+01:00 2012-05-18T12:41:06+02:00 espinoza http://cvpapers.com/forum/account.php?u=53

Believe it or not, corset style wedding dresses are essentially the most selected dresses amongst brides today. When fashion trends by no means stray away from the figure-flattering silhouette ...

plus size bridesmaid dresses fit into the most recent fashion sense. And thus nowadays we see a dazzling array of corset wedding dresses in fashion to become loved purple bridesmaid dresses by brides turn the chapel aisle into the runway. evening dresses for weddings About this page: sexy corset wedding black bridesmaid dresses dresses are hot everywhere in modern era, and in this write-up we'll talk about this kind of dress for a general thought of it and its style options.]]> http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=32#Comment_32 2010-06-30T02:02:39+02:00 2012-05-18T12:41:06+02:00 guest http://cvpapers.com/forum/account.php?u=2

My CVPR 2010 "Best Gestalt Paper" is Tracking the invisible: learning where the object may be"... "Awarding the invisible": the paper "Being Johan Malkovic", ...

"Awarding the invisible":
the paper "Being Johan Malkovic", even tough not yet available on the web, has good chance to win the ECCV 2010 "Best Gestalt Paper"]]> http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=31#Comment_31 2010-06-27T23:17:03+02:00 2012-05-18T12:41:06+02:00 overrider http://cvpapers.com/forum/account.php?u=6

>> http://www.vision.ucsd.edu/sites/default/files/gestalt.pdf Have someone recognized the papers on Fig. 6--7?

> http://www.vision.ucsd.edu/sites/default/files/gestalt.pdf

Have someone recognized the papers on Fig. 6--7?]]> http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=30#Comment_30 2010-06-27T19:06:01+02:00 2012-05-18T12:41:06+02:00 guest http://cvpapers.com/forum/account.php?u=2

Two papers: "Visual Tracking Decomposition" has impressive results (but it is surely CPU intensive) and "Efficient Hierarchical Graph-Based Video Segmentation".

"Visual Tracking Decomposition" has impressive results (but it is surely CPU intensive) and "Efficient Hierarchical Graph-Based Video Segmentation".]]> http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=29#Comment_29 2010-06-26T13:14:27+02:00 2012-05-18T12:41:06+02:00 maciej.gryka http://cvpapers.com/forum/account.php?u=9

These look interesting, thanks!

http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=28#Comment_28 2010-06-25T18:24:34+02:00 2012-05-18T12:41:06+02:00 cvpapers http://cvpapers.com/forum/account.php?u=1

Two papers available on CVPapers ;-) about line matching (ICCV 2009 & CVPR 2010): Wide-Baseline Image Matching Using Line Signatures (ICCV 2009) Line Matching Leveraged By Point Correspondences ...

Wide-Baseline Image Matching Using Line Signatures (ICCV 2009)
Line Matching Leveraged By Point Correspondences (CVPR 2010)

Hope it helps.]]> http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=27#Comment_27 2010-06-25T10:38:54+02:00 2012-05-18T12:41:06+02:00 Pierre http://cvpapers.com/forum/account.php?u=11

Did you have thought about line based matching ? I think contours will be the most repeatable things between the two types of images

I think contours will be the most repeatable things between the two types of images]]> http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=26#Comment_26 2010-06-25T10:38:26+02:00 2012-05-18T12:41:06+02:00 guest http://cvpapers.com/forum/account.php?u=2

Maybe with a preprocessing stage, using edge detector or LBP. Or features based on those operators...

http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=25#Comment_25 2010-06-25T10:17:34+02:00 2012-05-18T12:41:06+02:00 cvpapers http://cvpapers.com/forum/account.php?u=1

The correct link of this paper is http://vision.ucsd.edu/sites/default/files/gestalt.pdf You are right ! Interesting (and fun) paper.

http://vision.ucsd.edu/sites/default/files/gestalt.pdf
You are right ! Interesting (and fun) paper.]]> http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=24#Comment_24 2010-06-25T02:45:17+02:00 2012-05-18T12:41:06+02:00 guest http://cvpapers.com/forum/account.php?u=2

Not the best, but surely interesting... http://www.vision.ucsd.edu/sites/default/files/gestalt.pdf

http://www.vision.ucsd.edu/sites/default/files/gestalt.pdf]]> http://cvpapers.com/forum/comments.php?DiscussionID=9&Focus=23#Comment_23 2010-06-24T17:22:42+02:00 2012-05-18T12:41:06+02:00 cvpapers http://cvpapers.com/forum/account.php?u=1

Give your CVPR 2010 "Best Paper(s)"...

http://cvpapers.com/forum/comments.php?DiscussionID=8&Focus=22#Comment_22 2010-06-24T12:04:07+02:00 2012-05-18T12:41:06+02:00 maciej.gryka http://cvpapers.com/forum/account.php?u=9

What could be a good way to find correspondences between thermal and visual images? Examples here: visual, thermal SIFT doesn't work for that - it's great for visual-visual or thermal-thermal ...

visual, thermal

SIFT doesn't work for that - it's great for visual-visual or thermal-thermal correspondences, but not across types.

Thanks,
Maciek]]> http://cvpapers.com/forum/comments.php?DiscussionID=6&Focus=20#Comment_20 2010-06-22T22:55:33+02:00 2012-05-18T12:41:06+02:00 overrider http://cvpapers.com/forum/account.php?u=6

On the second point. Well, there is a [computer-vision] tag on SO. I think the stack-exchange engine is not appropriate for discussing papers. It assumes having 1-2 great answers to a precise ...

Well, there is a [computer-vision] tag on SO.
I think the stack-exchange engine is not appropriate for discussing papers. It assumes having 1-2 great answers to a precise question, which is not the case. Here we need somewhat for discussions like a forum.

Also look at the example of themachinelearningforum.org]]> http://cvpapers.com/forum/comments.php?DiscussionID=5&Focus=18#Comment_18 2010-06-22T22:30:23+02:00 2012-05-18T12:41:06+02:00 guest http://cvpapers.com/forum/account.php?u=2

computer vision rulez! I need it even to post this message. (I mean reCaptcha)

http://cvpapers.com/forum/comments.php?DiscussionID=6&Focus=15#Comment_15 2010-06-22T15:38:40+02:00 2012-05-18T12:41:06+02:00 mgryka http://cvpapers.com/forum/account.php?u=5

First of all thanks for organising all this - having so many interesting papers in one place is really cool. Secondly, it's great that you're adding the forum - I think a lot can be gained from ...

As for feedback - I don't want to suggest too much changes at the very start, but since you asked :)

• in terms of looks (nitpick really), the grey header on the top of the page could be smaller IMHO - it grabs quite a lot of attention from the content right now


• it'd be nice to have some more modern forum/discussion form, rather than the phpBB-style one. Obviously that would require scrapping everything so it's not a quick change, but something like Hackernews or StackOverflow would be really cool. Especially for more specific, focused topics like discussing a given paper. They both also do a great job of keeping the discussions on-topic and engaging community, but how much of that is due to form is arguable

I'll add more when something comes to my mind.

Thanks,
Maciek]]> http://cvpapers.com/forum/comments.php?DiscussionID=5&Focus=14#Comment_14 2010-06-22T15:02:52+02:00 2012-05-18T12:41:06+02:00 cvpapers http://cvpapers.com/forum/account.php?u=1

You can add file attachements to your post using the "Attachements" button at the bottom of the text area... See the result in this comment

"Attachements" button at the bottom of the text area...
See the result in this comment]]> http://cvpapers.com/forum/comments.php?DiscussionID=5&Focus=13#Comment_13 2010-06-22T15:00:51+02:00 2012-05-18T12:41:06+02:00 cvpapers http://cvpapers.com/forum/account.php?u=1

You can add HTML tag:Switch radio button, at the bottom of the text area, to HTMLSwitch this radio button to Text if you prefer to use text only

can add HTML tag:

• Switch radio button, at the bottom of the text area, to HTML
• Switch this radio button to Text if you prefer to use text only

]]>